From 5b8f0a480348774b5217963eba456d2a192884ca Mon Sep 17 00:00:00 2001
From: "devin-ai-integration[bot]"
 <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Date: Fri, 9 Oct 2026 16:07:13 -0700
Subject: [PATCH] fix(model_prices): registry audit, perplexity agent api
 models and claude-sonnet-4-5 200k context (#45647)

Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 ...odel_prices_and_context_window_backup.json | 57 ++++++++++++-------
 model_prices_and_context_window.json          | 57 ++++++++++++-------
 model_prices_and_context_window.schema.json   |  5 --
 3 files changed, 72 insertions(+), 47 deletions(-)

diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json
index 7b192dcd0d6..5aaab3f3544 100644
--- a/litellm/model_prices_and_context_window_backup.json
+++ b/litellm/model_prices_and_context_window_backup.json
@@ -2992,12 +2992,12 @@
     "anthropic.claude-sonnet-4-5-20250929-v1:0": {
         "cache_creation_input_token_cost": 3.75e-06,
         "cache_creation_input_token_cost_above_1hr": 6e-06,
+        "cache_creation_input_token_cost_above_1hr_above_200k_tokens": 1.2e-05,
         "cache_read_input_token_cost": 3e-07,
         "input_cost_per_token": 3e-06,
         "input_cost_per_token_above_200k_tokens": 6e-06,
         "output_cost_per_token_above_200k_tokens": 2.25e-05,
         "cache_creation_input_token_cost_above_200k_tokens": 7.5e-06,
-        "cache_creation_input_token_cost_above_1hr_above_200k_tokens": 1.2e-05,
         "cache_read_input_token_cost_above_200k_tokens": 6e-07,
         "deprecation_date": "2027-04-08",
         "litellm_provider": "bedrock_converse",
@@ -15256,27 +15256,18 @@
     "claude-sonnet-4-5": {
         "cache_creation_input_token_cost": 3.75e-06,
         "cache_creation_input_token_cost_above_1hr": 6e-06,
-        "cache_creation_input_token_cost_above_1hr_above_200k_tokens": 1.2e-05,
         "cache_read_input_token_cost": 3e-07,
         "input_cost_per_token": 3e-06,
-        "input_cost_per_token_above_200k_tokens": 6e-06,
-        "output_cost_per_token_above_200k_tokens": 2.25e-05,
-        "cache_creation_input_token_cost_above_200k_tokens": 7.5e-06,
-        "cache_creation_input_token_cost_above_200k_tokens_batches": 3.75e-06,
         "cache_creation_input_token_cost_batches": 1.875e-06,
-        "cache_read_input_token_cost_above_200k_tokens": 6e-07,
-        "cache_read_input_token_cost_above_200k_tokens_batches": 3e-07,
         "cache_read_input_token_cost_batches": 1.5e-07,
         "deprecation_date": "2026-11-30",
-        "input_cost_per_token_above_200k_tokens_batches": 3e-06,
         "input_cost_per_token_batches": 1.5e-06,
         "litellm_provider": "anthropic",
-        "max_input_tokens": 1000000,
+        "max_input_tokens": 200000,
         "max_output_tokens": 64000,
         "max_tokens": 64000,
         "mode": "chat",
         "output_cost_per_token": 1.5e-05,
-        "output_cost_per_token_above_200k_tokens_batches": 1.125e-05,
         "output_cost_per_token_batches": 7.5e-06,
         "search_context_cost_per_query": {
             "search_context_size_high": 0.01,
@@ -15299,27 +15290,18 @@
     "claude-sonnet-4-5-20250929": {
         "cache_creation_input_token_cost": 3.75e-06,
         "cache_creation_input_token_cost_above_1hr": 6e-06,
-        "cache_creation_input_token_cost_above_1hr_above_200k_tokens": 1.2e-05,
         "cache_read_input_token_cost": 3e-07,
         "input_cost_per_token": 3e-06,
-        "input_cost_per_token_above_200k_tokens": 6e-06,
-        "output_cost_per_token_above_200k_tokens": 2.25e-05,
-        "cache_creation_input_token_cost_above_200k_tokens": 7.5e-06,
-        "cache_creation_input_token_cost_above_200k_tokens_batches": 3.75e-06,
         "cache_creation_input_token_cost_batches": 1.875e-06,
-        "cache_read_input_token_cost_above_200k_tokens": 6e-07,
-        "cache_read_input_token_cost_above_200k_tokens_batches": 3e-07,
         "cache_read_input_token_cost_batches": 1.5e-07,
         "deprecation_date": "2026-11-30",
-        "input_cost_per_token_above_200k_tokens_batches": 3e-06,
         "input_cost_per_token_batches": 1.5e-06,
         "litellm_provider": "anthropic",
-        "max_input_tokens": 1000000,
+        "max_input_tokens": 200000,
         "max_output_tokens": 64000,
         "max_tokens": 64000,
         "mode": "chat",
         "output_cost_per_token": 1.5e-05,
-        "output_cost_per_token_above_200k_tokens_batches": 1.125e-05,
         "output_cost_per_token_batches": 7.5e-06,
         "search_context_cost_per_query": {
             "search_context_size_high": 0.01,
@@ -43298,6 +43280,14 @@
         "supports_reasoning": true,
         "supports_function_calling": true
     },
+    "perplexity/perplexity/deepseek-v4-pro-0813": {
+        "litellm_provider": "perplexity",
+        "mode": "responses",
+        "input_cost_per_token": 1.32e-06,
+        "output_cost_per_token": 3.96e-06,
+        "cache_read_input_token_cost": 4.4e-08,
+        "source": "https://docs.perplexity.ai/docs/agent-api/models"
+    },
     "perplexity/perplexity/glm-5.2": {
         "cache_read_input_token_cost": 1.4e-07,
         "input_cost_per_token": 1.4e-06,
@@ -65168,6 +65158,22 @@
         "cache_read_input_token_cost": 1.25e-07,
         "source": "https://docs.perplexity.ai/docs/agent-api/models"
     },
+    "perplexity/openai/gpt-5-nano": {
+        "litellm_provider": "perplexity",
+        "mode": "responses",
+        "input_cost_per_token": 5e-08,
+        "output_cost_per_token": 4e-07,
+        "cache_read_input_token_cost": 5e-09,
+        "source": "https://docs.perplexity.ai/docs/agent-api/models"
+    },
+    "perplexity/openai/gpt-5.3-codex": {
+        "litellm_provider": "perplexity",
+        "mode": "responses",
+        "input_cost_per_token": 3.5e-06,
+        "output_cost_per_token": 2.8e-05,
+        "cache_read_input_token_cost": 3.5e-07,
+        "source": "https://docs.perplexity.ai/docs/agent-api/models"
+    },
     "perplexity/google/gemini-3.1-pro-preview": {
         "litellm_provider": "perplexity",
         "mode": "responses",
@@ -70074,6 +70080,15 @@
         "cache_read_input_token_cost_above_272k_tokens": 4e-07,
         "source": "https://docs.perplexity.ai/docs/agent-api/models"
     },
+    "perplexity/openai/gpt-6-astra": {
+        "litellm_provider": "perplexity",
+        "mode": "responses",
+        "input_cost_per_token": 1e-05,
+        "output_cost_per_token": 5e-05,
+        "cache_read_input_token_cost": 1e-06,
+        "cache_creation_input_token_cost": 1.25e-05,
+        "source": "https://docs.perplexity.ai/docs/agent-api/models"
+    },
     "perplexity/openai/gpt-6-luna": {
         "litellm_provider": "perplexity",
         "mode": "responses",
diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json
index 7b192dcd0d6..5aaab3f3544 100644
--- a/model_prices_and_context_window.json
+++ b/model_prices_and_context_window.json
@@ -2992,12 +2992,12 @@
     "anthropic.claude-sonnet-4-5-20250929-v1:0": {
         "cache_creation_input_token_cost": 3.75e-06,
         "cache_creation_input_token_cost_above_1hr": 6e-06,
+        "cache_creation_input_token_cost_above_1hr_above_200k_tokens": 1.2e-05,
         "cache_read_input_token_cost": 3e-07,
         "input_cost_per_token": 3e-06,
         "input_cost_per_token_above_200k_tokens": 6e-06,
         "output_cost_per_token_above_200k_tokens": 2.25e-05,
         "cache_creation_input_token_cost_above_200k_tokens": 7.5e-06,
-        "cache_creation_input_token_cost_above_1hr_above_200k_tokens": 1.2e-05,
         "cache_read_input_token_cost_above_200k_tokens": 6e-07,
         "deprecation_date": "2027-04-08",
         "litellm_provider": "bedrock_converse",
@@ -15256,27 +15256,18 @@
     "claude-sonnet-4-5": {
         "cache_creation_input_token_cost": 3.75e-06,
         "cache_creation_input_token_cost_above_1hr": 6e-06,
-        "cache_creation_input_token_cost_above_1hr_above_200k_tokens": 1.2e-05,
         "cache_read_input_token_cost": 3e-07,
         "input_cost_per_token": 3e-06,
-        "input_cost_per_token_above_200k_tokens": 6e-06,
-        "output_cost_per_token_above_200k_tokens": 2.25e-05,
-        "cache_creation_input_token_cost_above_200k_tokens": 7.5e-06,
-        "cache_creation_input_token_cost_above_200k_tokens_batches": 3.75e-06,
         "cache_creation_input_token_cost_batches": 1.875e-06,
-        "cache_read_input_token_cost_above_200k_tokens": 6e-07,
-        "cache_read_input_token_cost_above_200k_tokens_batches": 3e-07,
         "cache_read_input_token_cost_batches": 1.5e-07,
         "deprecation_date": "2026-11-30",
-        "input_cost_per_token_above_200k_tokens_batches": 3e-06,
         "input_cost_per_token_batches": 1.5e-06,
         "litellm_provider": "anthropic",
-        "max_input_tokens": 1000000,
+        "max_input_tokens": 200000,
         "max_output_tokens": 64000,
         "max_tokens": 64000,
         "mode": "chat",
         "output_cost_per_token": 1.5e-05,
-        "output_cost_per_token_above_200k_tokens_batches": 1.125e-05,
         "output_cost_per_token_batches": 7.5e-06,
         "search_context_cost_per_query": {
             "search_context_size_high": 0.01,
@@ -15299,27 +15290,18 @@
     "claude-sonnet-4-5-20250929": {
         "cache_creation_input_token_cost": 3.75e-06,
         "cache_creation_input_token_cost_above_1hr": 6e-06,
-        "cache_creation_input_token_cost_above_1hr_above_200k_tokens": 1.2e-05,
         "cache_read_input_token_cost": 3e-07,
         "input_cost_per_token": 3e-06,
-        "input_cost_per_token_above_200k_tokens": 6e-06,
-        "output_cost_per_token_above_200k_tokens": 2.25e-05,
-        "cache_creation_input_token_cost_above_200k_tokens": 7.5e-06,
-        "cache_creation_input_token_cost_above_200k_tokens_batches": 3.75e-06,
         "cache_creation_input_token_cost_batches": 1.875e-06,
-        "cache_read_input_token_cost_above_200k_tokens": 6e-07,
-        "cache_read_input_token_cost_above_200k_tokens_batches": 3e-07,
         "cache_read_input_token_cost_batches": 1.5e-07,
         "deprecation_date": "2026-11-30",
-        "input_cost_per_token_above_200k_tokens_batches": 3e-06,
         "input_cost_per_token_batches": 1.5e-06,
         "litellm_provider": "anthropic",
-        "max_input_tokens": 1000000,
+        "max_input_tokens": 200000,
         "max_output_tokens": 64000,
         "max_tokens": 64000,
         "mode": "chat",
         "output_cost_per_token": 1.5e-05,
-        "output_cost_per_token_above_200k_tokens_batches": 1.125e-05,
         "output_cost_per_token_batches": 7.5e-06,
         "search_context_cost_per_query": {
             "search_context_size_high": 0.01,
@@ -43298,6 +43280,14 @@
         "supports_reasoning": true,
         "supports_function_calling": true
     },
+    "perplexity/perplexity/deepseek-v4-pro-0813": {
+        "litellm_provider": "perplexity",
+        "mode": "responses",
+        "input_cost_per_token": 1.32e-06,
+        "output_cost_per_token": 3.96e-06,
+        "cache_read_input_token_cost": 4.4e-08,
+        "source": "https://docs.perplexity.ai/docs/agent-api/models"
+    },
     "perplexity/perplexity/glm-5.2": {
         "cache_read_input_token_cost": 1.4e-07,
         "input_cost_per_token": 1.4e-06,
@@ -65168,6 +65158,22 @@
         "cache_read_input_token_cost": 1.25e-07,
         "source": "https://docs.perplexity.ai/docs/agent-api/models"
     },
+    "perplexity/openai/gpt-5-nano": {
+        "litellm_provider": "perplexity",
+        "mode": "responses",
+        "input_cost_per_token": 5e-08,
+        "output_cost_per_token": 4e-07,
+        "cache_read_input_token_cost": 5e-09,
+        "source": "https://docs.perplexity.ai/docs/agent-api/models"
+    },
+    "perplexity/openai/gpt-5.3-codex": {
+        "litellm_provider": "perplexity",
+        "mode": "responses",
+        "input_cost_per_token": 3.5e-06,
+        "output_cost_per_token": 2.8e-05,
+        "cache_read_input_token_cost": 3.5e-07,
+        "source": "https://docs.perplexity.ai/docs/agent-api/models"
+    },
     "perplexity/google/gemini-3.1-pro-preview": {
         "litellm_provider": "perplexity",
         "mode": "responses",
@@ -70074,6 +70080,15 @@
         "cache_read_input_token_cost_above_272k_tokens": 4e-07,
         "source": "https://docs.perplexity.ai/docs/agent-api/models"
     },
+    "perplexity/openai/gpt-6-astra": {
+        "litellm_provider": "perplexity",
+        "mode": "responses",
+        "input_cost_per_token": 1e-05,
+        "output_cost_per_token": 5e-05,
+        "cache_read_input_token_cost": 1e-06,
+        "cache_creation_input_token_cost": 1.25e-05,
+        "source": "https://docs.perplexity.ai/docs/agent-api/models"
+    },
     "perplexity/openai/gpt-6-luna": {
         "litellm_provider": "perplexity",
         "mode": "responses",
diff --git a/model_prices_and_context_window.schema.json b/model_prices_and_context_window.schema.json
index 11a65b81662..7f58501eed5 100644
--- a/model_prices_and_context_window.schema.json
+++ b/model_prices_and_context_window.schema.json
@@ -113,11 +113,6 @@
           "minimum": 0,
           "description": "Rate applied once the prompt exceeds the token threshold in the field name."
         },
-        "cache_creation_input_token_cost_above_200k_tokens_batches": {
-          "type": "number",
-          "minimum": 0,
-          "description": "Rate applied once the prompt exceeds the token threshold in the field name."
-        },
         "cache_creation_input_token_cost_above_272k_tokens": {
           "type": "number",
           "minimum": 0,